What model Deskpro is that? Could be using another CMOS battery that they had inside their machines instead of the CR2032's. Or, it could be a rechargeable lithium ML2032. Worst case scenario, a cold, cracked, or bad solder joints under the battery clip itself. Check for corrosion and clean it off with DeoxIt since corrosion can cause the battery to die faster than usual, and re-flash the BIOS chip as well since that might be another issue.

Try to put a multimeter between the battery and the board.
Tell us how many milliamperes your board draws from the battery.

It should be a few microamperes.

I built a computer for my aunt that had an odd problem - whenever I would move it, it would clear the CMOS. I finally discovered that I had a standoff that didn't have a corresponding hole in the motherboard, and when it was jostled it would ground out the motherboard, clearing the CMOS memory. If it had been a little tighter it would have quickly discharged the battery, so I would make double sure you don't have an unnecessary standoff between the case & the motherboard.

bjwil1991 wrote:

Worst case scenario, a cold, cracked, or bad solder joints under the battery clip itself. Check for corrosion and clean it off with DeoxIt since corrosion can cause the battery to die faster than usual, and re-flash the BIOS chip as well since that might be another issue.

How can any of the above "cause the battery to die faster than usual"?

Any problem with the solders can increase the resistance in ohms, so it will draw more power from the battery. Buggy BIOS can left some devices from the board awake, but I guess that would have been noted long before.

Buggy BIOS can certainly cause this. GVC's FR520 was notorious for it, until you updated to the last BIOS for the board, it ate the CR-2032s faster than you could put them in there. Never heard of a specific issue with Compaq systems, but worth checking if there is a newer BIOS that specifically fixes this kind of issue...